Sign in with
Sign up | Sign in
Your question

QX6700 @ 3.4 ghz Need to go higher, any advice ?

Last response: in Overclocking
Share
December 15, 2006 4:00:36 AM

I got my Qx6700 to 3.4 fully stable with the following setting
Cpu freq: 340
Vcore: 1.378
Multiplier: 10 (I can't go any higher, qx6700 limit 10 ????)
Dram timing by SPD
Dram freq: 680
Dram voltage 2.0V
I am using swiftech H220 water cooling, currently iddle at 40C and full load all 4 cores 100% at 58C
My spec: P5W DH, 4g of DDR2 800, 480W Antec PSU, and a low end graphic card (I don't need much graphic)
I tried 3.6 with vcore 1.4375 but it crash after 10 mins full load and Core Temp reached 64C
Some one have experience with Qx6700, please give me some advice. My goal is 4.0 ghz
December 15, 2006 4:34:14 AM

Do you use that for gaming?
December 15, 2006 4:43:25 AM

Quote:
I got my Qx6700 to 3.4 fully stable with the following setting
Cpu freq: 340
Vcore: 1.378
Multiplier: 10 (I can't go any higher, qx6700 limit 10 ????)
Dram timing by SPD
Dram freq: 680
Dram voltage 2.0V
I am using swiftech H220 water cooling, currently iddle at 40C and full load all 4 cores 100% at 58C
My spec: P5W DH, 4g of DDR2 800, 480W Antec PSU, and a low end graphic card (I don't need much graphic)
I tried 3.6 with vcore 1.4375 but it crash after 10 mins full load and Core Temp reached 64C
Some one have experience with Qx6700, please give me some advice. My goal is 4.0 ghz


You will not get a Qx6700 to 4 GHz on water. It runs insanely hot OC'd. The phase people are having to retune their units to deal with running this processor full out. You are getting 58C @ 1.378 using water. Think about that. You will need closing in on 1.6 vcore to hit 4 GHZ in all probability if your cpu clocks like many people are seeing. Now extrapolate from 58/1.378 to X/1.6 (and it's not a linear rise and its not in your favor). Further, you could be getting instability from doing this with 4 gig of RAM. Next, what Antec PSU? You want solidly stable heavy duty quality power for high overclocks on Qx6700 and 4 gig of RAM. 480W is low for this CPU, IMO. Plus it's Antec, which isn't top tier in any event.
Related resources
December 15, 2006 8:01:15 PM

I use it for heavy computation, not for gaming. I do think that my PSU is weak, however, do you think the psu is a real factor at this point, all power flow is stable and sufficient when i check them. But the temp and voltage is the problem I have. I do not know any other cooling beside water cooling. Is there a good cooling trick that I don't know about ?
December 15, 2006 9:25:52 PM

Quote:
I use it for heavy computation, not for gaming. I do think that my PSU is weak, however, do you think the psu is a real factor at this point, all power flow is stable and sufficient when i check them. But the temp and voltage is the problem I have. I do not know any other cooling beside water cooling. Is there a good cooling trick that I don't know about ?


Unless you want to go TEC, phase change or chillers you are ultimately limited in your overclock. I'm not saying you should move to these advanced/riskier cooling techniques. You have to do your own risk vs. reward vs. cost analysis.

As to your PSU, QX6700 is an $1100+ processor. The quality of your PSU should be on par, especially if you are going to OC it substantially. Where does your PSU fall on this list?
http://www.xtremesystems.org/forums/showthread.php?t=10...

See the last few pages of this thread regarding some of the difficulty people are seeing even with phase cooling for OC'd QX6700:
http://www.xtremesystems.org/forums/showthread.php?t=92...

Also, what kind of computation? Where does the stress lie? RAM bound? Paging to hard drive a lot?
December 17, 2006 6:32:19 AM

Swiftech will provide enough cooling to reach 4Ghz, dont worry about that.

Raise vMCH to the max. the NB will run very hot after this of course, thats why I strongly advice you to slap a small fan on it (or even water cool it).
Run the advertised voltage on the RAM but use 5-5-5-15 2T timings.
Vcore shouldnt go over 1.45v to keep temps acceptable.
Give vFSB some help with the voltage and raise vICH (SB) by just a notch.

Remember that you have a Tjunction of 100c, so youve got more headroom when it comes to temperatures than a C2D.
a b à CPUs
a b K Overclocking
December 17, 2006 8:59:37 AM

10X multiplier is the max for the Qx6700.

To date, the only member of the C2 family that you can up the multiplier on is the x6800.

Everything I have heard tells me that you can not push your cpu verry far.
Proper cooling and power supply to 4 cores in that small of a package make it next to impossible to reach the same levels of OC'ing as it's dual core brethren.

You have already gotten substansial increases out of your investment.
It is running at ~127.5% of stock already which is verry respectiable.

With the Qx6700 costing bout $1600 us, I would recomend erroring on the side of caution and leaving it were it is.
December 17, 2006 2:48:26 PM

Quote:
Swiftech will provide enough cooling to reach 4Ghz, dont worry about that.

Raise vMCH to the max. the NB will run very hot after this of course, thats why I strongly advice you to slap a small fan on it (or even water cool it).
Run the advertised voltage on the RAM but use 5-5-5-15 2T timings.
Vcore shouldnt go over 1.45v to keep temps acceptable.
Give vFSB some help with the voltage and raise vICH (SB) by just a notch.

Remember that you have a Tjunction of 100c, so youve got more headroom when it comes to temperatures than a C2D.


He has vcore of 1.378 @ 3.4GHz. Going to 1.45v will not get him to 4 GHz on this processor. Water cooling, unless you have a golden processor, will not run this quad core at 4 GHz in any event. It's just too hot. Espcially since it seems he wants it there for daily use.
December 17, 2006 2:52:01 PM

Quote:
With the Qx6700 costing bout $1600 us, I would recomend erroring on the side of caution and leaving it were it is.


You are correct that it is best to be cautious, but your price sounds like the newegg one (which is insane). That proc can be had for $1200 retail here:
http://www.tankguys.biz/intel-quad-core-qx6700-kentsfie...

And I've seen it for just over $1k occassionally elsewhere. Very careful shopping is needed on this proc.
December 17, 2006 6:17:48 PM

Blah Blah PFF, you just wait and see.
December 17, 2006 6:30:38 PM

Quote:
Blah Blah PFF, you just wait and see.


What is "PFF", the Pakistan Football Federation?
December 17, 2006 7:55:07 PM

I also somewhat doubt he can hit 4ghz if he is already at 58C under load with the water kit at only 3.4ghz. Xtremesystems had a thread when coolaler first got a QX6700 and under a cascade he could only hit 4.0ghz on one of the chips he tested. Remember that with 4 cores it will be limited by it's weakest core and your odds of getting a "golden" cpu decline greatly when sticking more cores on the die.
December 17, 2006 8:35:28 PM

Quote:
I also somewhat doubt he can hit 4ghz if he is already at 58C under load with the water kit at only 3.4ghz. Xtremesystems had a thread when coolaler first got a QX6700 and under a cascade he could only hit 4.0ghz on one of the chips he tested. Remember that with 4 cores it will be limited by it's weakest core and your odds of getting a "golden" cpu decline greatly when sticking more cores on the die.

Yup.
December 17, 2006 9:16:10 PM

Temperatures dont concern me, since its clear that this wont be a 24/7 OC.
And Coolaler tested an ES, the ES for C2Q were poorly produced in comparison to the few retail samples in the market. In any case I think the MB will be the limited factor, the P5W is knowned for having a very weak NBCC.
December 17, 2006 10:26:10 PM

Quote:
Temperatures dont concern me, since its clear that this wont be a 24/7 OC.
And Coolaler tested an ES, the ES for C2Q were poorly produced in comparison to the few retail samples in the market. In any case I think the MB will be the limited factor, the P5W is knowned for having a very weak NBCC.


He said he wants it for heavy computation. That implies extended periods of heavy load to me.
Temperatures may not concern you, but I suspect they concern the guy with the $1200 cpu you are advising.
If he's 58C at only 3.4 GHz on water what do you think he will be pulling at 4 GHz? There is usually a huge vcore differential between these performance levels.
December 17, 2006 10:40:26 PM

The 4Ghz wont be 24/7 thats for sure, but any speed he achieves stabily at my recommended settings could be.

Quote:
Temperatures may not concern you, but I suspect they concern the guy with the $1200 cpu you are advising.
Quote:
Remember that you have a Tjunction of 100c, so youve got more headroom when it comes to temperatures than a C2D.
:lol: 
December 17, 2006 10:52:41 PM

First let’s see where your thermal limits are.
Set the Vcore to 1.55 absolute max allowed by Intel.
Run up the FSB to 366X10 make sure hyper path 3 is disabled in the chipset section of the bios menu.
Run prime 95 on all four cores and see if your temps reach no higher than 61.4C lets call it 62C.
If your stable and temps are in limits you can bring up the FSB by small increments until you achieve your max Oc on water.
If you have this pump http://www.swiftnets.com/products/mcp655.asp
Make sure the potentiometer on the back of the mcp655 pump is maxed out if you have the latest revised pump. (Fully clockwise)
Watch the flow thru the reservoir to ensure your flow is actually increasing.
This board is a little weak on the high end of the FSB so leave the multiplier at 10
Some of these boards have a hard time at around 300fsb so up the Northbridge volts to 1.45 or 1.55 and go straight to 366 the same happens around 380 to 400.
You may have to do a series of resets to stabilize the board so write down the settings because the board freeks out when you push it until you hit the sweet spot.
if your able to keep your temps low and are not afraid of Vcore you can push it further but if you hit your thermal limits at 366 @ 1.55 of cant post you have to approach this from a different angle
December 17, 2006 11:16:02 PM

Quote:
The 4Ghz wont be 24/7 thats for sure, but any speed he achieves stabily at my recommended settings could be.

Again, he said he wants it for heavy computation. That implies extended periods of heavy load to me. You think otherwise why? Do you interpret his statement:
Quote:
I use it for heavy computation, not for gaming.

as Spi 1M only or something?

I don't see people running their primary computational activities (i.e. whatever daily WORK they are doing) at 4 GHz with water on the Quads. If you do, please link to those people and what it is they are using and doing.
December 17, 2006 11:44:43 PM

Quote:
Run prime 95 on all four cores and see if your temps reach no higher than 61.4C lets call it 62C.
Gaah, no. That measurement was taken on the surface of the IHS on C2Ds, the temperatures reflected by TAT and CoreTemp are equations from the thermal probe within the CPU, therefore they WILL be higher.
100c (Tjunction) is the actual temperature to avoid at all times.
This article will cerainly help.
December 17, 2006 11:55:03 PM

Tjunction is meltdown my friend.
the only temp you can monitor is with core temp and the asus software pc Probe II along with a few other generics.
the Temps to monitor are as stated.
the cpu will handle anything over 80C
where in theory it will shut down at its set point.
you can not monitor Tjunction
read the data sheets to understand my reply .
this is where Intel sets its thermal limits and I dont think he has a thermal probe a mica cutter and the fluke to measure the actual temp at the heat spreader between the waterblock and cpu
December 18, 2006 12:28:12 AM

Tjunction is not meant to be monitored, its as you said "the meltdown" for every CPU, its a value setted by Intel (even that if youre knowable in C++ you can change it). But C2s throttle at such temp to prevent damage.
It seems to me that we are on the same idea, did you read the article?
December 18, 2006 12:36:37 AM

Doesn't he have a QX6700, is that not an Extreme processor which would include an unlocked multiplier..., just wondering cause i thought all Extreme editions were unlocked? not that would help much,

and one last thing, why do you need to increase the speed of your processor since it is pretty much one of the most powerful single socket setups around.., i mean thats like saying my Bugatti Veyron isn't fast enough i need 300 more hp, you've already beaten everyone else and then some, how much more power do you need!? :p  :p  :p  Take it easy Captain Kirk, "scotty, i need more power" lol
December 18, 2006 12:37:22 AM

yes I read the article and I want no part of your argument . :wink:
You cant change a binned chip or we would all own X6800's
but if you figure it out let me know because I sure need to unlock upwards 8)
December 18, 2006 12:42:41 AM

Quote:
Tjunction is not meant to be monitored, its as you said "the meltdown" for every CPU, its a value setted by Intel (even that if youre knowable in C++ you can change it).

You can change it? What are you talking about? If you are talking reprogramming core temp...who would care? If you are talking changing the property of the chip itself... :roll:
I know my C++. Explain to me in your own words how to change it.
December 18, 2006 12:52:02 AM

Its an Extreme Edition processor, and from the past all EE processor's came with unlocked multipliers, that was part of the reason why they demanded the high price. Which is why i question why you say you cannot change yours

When you say we would all own if we could change the multi upwards we would own X6800s, thats not true, as processor's are binned so they can run at that speed. lucky for us a E6300 can reach 6800 speeds easily, that usually is not the case which is why C2D is such a good chip.

In other words a chip binned to be a E6300 then overclocked(by the multiplier) to run at 2.99Ghz would not run as cool as a X6800 running at its stock speed of 2.99Ghz...this doesn't mean much but i was just pointing it out.
December 18, 2006 12:55:45 AM

That was my sick sarcasm
:wink:
December 18, 2006 1:14:46 AM

and yes I believe the multiplier is unlocked from 6 to 20
December 18, 2006 1:50:03 AM

You can ask the author to give you the source codes so you can re-program CoreTemp with manually/auto adjustable Tcasemax.
Dont mess with Tjunction because that WILL fry chips.
December 20, 2006 2:50:59 AM

I'm having the same problem. QX6700 on a P5W-DH. Won't allow the multiplier to be set over 10. All of the reviews of the QX said it has a totally unlocked multiplier and many of them show overclocks with the multiplier as high as 17.

I've managed to get my system stable at 3300mhz 'the old fashioned way' by tweaking the clock, but I think I could go higher and be more stable if I could leave the FSB alone. It is pretty irritating to expect a part to have an unlocked multiplier and then discover it ain't so.

Any ideas?
December 20, 2006 3:15:13 AM

do you have the latest bios revision ?
I believe that may have been resolved(was with the P5B)
but I cant find the link that supports that .
December 20, 2006 7:56:30 PM

I have a QX6700 CPU, EVGA NF680i MB, Corsair XMS2 800mhz 4-4-4-4 ram, silverstone ST75f and a decent case, the Temjin TJ09.

using air cooling, how far do u think I could clock my CPU to be used everyday under high loads? - ie. graphics rendering and simulation.

I know I have OC potential here, and im guessing I can raise the CPU speed a bit over its 2.66 speed.

Dont mind buying a few extra fans etc..
December 20, 2006 8:58:07 PM

Should be able to easily clock it to 3.0Ghz, after that will require a little more tweaking and keeping a close eye on temperatures and such. Once you clock it to 3.0Ghz you will have a smoking system, it can but i wouldn't bother going above that.
December 20, 2006 9:24:16 PM

Quote:
I have a QX6700 CPU, EVGA NF680i MB, Corsair XMS2 800mhz 4-4-4-4 ram, silverstone ST75f and a decent case, the Temjin TJ09.

using air cooling, how far do u think I could clock my CPU to be used everyday under high loads? - ie. graphics rendering and simulation.

I know I have OC potential here, and im guessing I can raise the CPU speed a bit over its 2.66 speed.

Dont mind buying a few extra fans etc..

I'm getting 3.3 on the Qx6700, air cooled, nothing special, only the most minor voltage tweeks (Vcore @ 1.375v, Vdim @ 2.2v). No extreme temp problems at all. As I said before, I had to change clock timing to do this because my mobo won't let me go above 10 on the multiplier. It gets unstable going higher than this, apparently due to memory timing problems - that's why I think I could do better if I could leave the clock alone and just deal with the multiplier.
December 20, 2006 11:17:27 PM

realisticly if you water cool 3.6Ghz 24/7
on air 3.4 but 3.2 should be a goal for stability and lower Vcore
with a good cooler 3.4
that ram will help you maintain a good overclock as long as there is good case airflow.
Anonymous
December 21, 2006 2:30:05 AM

Use some liquid nitrogen to cool the cpu down.
December 21, 2006 3:25:45 AM

Quote:
10X multiplier is the max for the Qx6700.

To date, the only member of the C2 family that you can up the multiplier on is the x6800.

Everything I have heard tells me that you can not push your cpu verry far.
Proper cooling and power supply to 4 cores in that small of a package make it next to impossible to reach the same levels of OC'ing as it's dual core brethren.

You have already gotten substansial increases out of your investment.
It is running at ~127.5% of stock already which is verry respectiable.

With the Qx6700 costing bout $1600 us, I would recomend erroring on the side of caution and leaving it were it is.
Because the QX6700 is an "Extreme Edition" chip, the multiplier should be fully unlocked.
December 21, 2006 12:08:13 PM

Quote:
10X multiplier is the max for the Qx6700.

To date, the only member of the C2 family that you can up the multiplier on is the x6800.

Everything I have heard tells me that you can not push your cpu verry far.
Proper cooling and power supply to 4 cores in that small of a package make it next to impossible to reach the same levels of OC'ing as it's dual core brethren.

You have already gotten substansial increases out of your investment.
It is running at ~127.5% of stock already which is verry respectiable.

With the Qx6700 costing bout $1600 us, I would recomend erroring on the side of caution and leaving it were it is.
Because the QX6700 is an "Extreme Edition" chip, the multiplier should be fully unlocked.

Your right

The multiplier is unlocked and I dont know why some people state that its not.
http://www.pcw.co.uk/personal-computer-world/hardware/2...
I would post a hundred related articles but why.

any way I originally said to leave the multiplier at 10X to find the thermal limits of the chip before he overclocks.
somehow it became a dead end thread full of pointless speculations.
December 21, 2006 2:32:31 PM

The problem is not that the multiplier is locked - the problem is that you can't set the (unlocked) multiplier above 10 when using the P5W-DH mobo.

The question is: do I need to look for a new board or is there something I am overlooking?

Of course, now that i've got it stable on air 7/24 at 3.3 w/out changing the multiplier maybe I don't really care anymore...
December 21, 2006 3:11:07 PM

Quote:
I got my Qx6700 to 3.4 fully stable with the following setting
Cpu freq: 340
Vcore: 1.378
Multiplier: 10 (I can't go any higher, qx6700 limit 10 ????)
Dram timing by SPD
Dram freq: 680
Dram voltage 2.0V
I am using swiftech H220 water cooling, currently iddle at 40C and full load all 4 cores 100% at 58C
My spec: P5W DH, 4g of DDR2 800, 480W Antec PSU, and a low end graphic card (I don't need much graphic)
I tried 3.6 with vcore 1.4375 but it crash after 10 mins full load and Core Temp reached 64C
Some one have experience with Qx6700, please give me some advice. My goal is 4.0 ghz

Take your mobo, CPU, and ect... which has none moving parts and place it in a tank with a large amount of cooking oil. Cool the cooking oil in the tank close to a gel. This should make your mobo more stable and or able to OC more. The cooled cooking oil should help cool your CPU as well. Note: This will only work if you GPU is water cooled also.
December 22, 2006 12:13:32 AM

Flashed 1707 yesterday. No change, still can't modify multiplier above 10.

Getting bored with it. The overclock is running stable and I'm back to editing/rendering (which is why I wanted this monster to begin with). Maybe I'll get back to it after new years...
December 22, 2006 1:50:15 AM

Quote:
[...]
Of course, now that i've got it stable on air 7/24 at 3.3 w/out changing the multiplier maybe I don't really care anymore...

Stable on air 7/24 at 3.30GHz... Now THAT'S what I'm talkin about.

Indeed why would you care, my man?

Your system is a knuckle dragging crunch monster!!!

With 4GB of fast RAM and four blazing cores? Your system is a perfect VMware server. With your system running VMware Server 1.01 on top of Windows Server 2003 you could run between 8 and 16 fairly powerful virtual systems. With your setup right now you could run two Win2k3 SP1 servers configured as Domain Controllers, four WinXP SP2 workstations, and two Linux server boxes, no problem whatsoever.

I just noticed you be rendering and editing, but the same goes. You literally have a faster, more powerful system than you could buy anywhere without going to 8 cores. You're smart to stop and enjoy the smoothness.

Although I think your PSU may be a weak link if you really push the system hard for continuous operation. If you have any problems that is likely to be your culprit.

Grip it and rip it.
December 24, 2006 5:50:09 PM

I recently change my psu to 700W OCZ. I am currenty running at 3.5Ghz stable. With the following setting:

Bios 1707
CPU Freq: 350
Multiplier: 10 (can't change higher)
DRAM Freq: DDR2-700MHz
PCI Exp. Freq: 100
PCI Clock Syn Mode: 33.33Mhz
Memory Voltage: Auto
Cpu Vcore: 1.4375V
FsB: 1.4V
MCH: 1.65V
ICH: 1.2V
DRAM Timing 5-5-5 - 15
Hyper Path 3: Enable (the solfware that I use somehow perform better when HP3 enable)
Dram throttling: enable
Under full load all 4 cores 100% my Coretemp read: 60C max, and idle at 42C
I only use my pc full load continiously 2 hrs max.
December 24, 2006 6:02:35 PM

You’re in the right direction.
With that chip you couldn’t ask for more for processing power.
Of course with a little more effort and luck on your side I’m confident you can get better results than that.
The real point is you’re happy with what you have accomplished and should be.
Good job :wink:
Merry Xmas and a safe and happy New Year.
December 24, 2006 6:12:08 PM

That's very strange; have you contacted Intel about this? All Extreme Edition chips are supposed to come with a fully unlocked multiplier.
December 24, 2006 6:24:14 PM

I thought the same thing .
The P5B Deluxe supports this and others so there seems to be an Asus problem here .
or maybe an overlooked setting in bios.
I cant confirm this until after New Years when I return to work .(if I remember) :lol: 
December 24, 2006 6:45:33 PM

Its a shame you spent out so much on a CPU and dont have the motherboard to take it to its limit!

Heres another link
http://www.legitreviews.com/article/395/4/

Stable at 3.667ghz with a 1333mhz fsb on 11x on watercooling. Yes please!
December 24, 2006 7:14:16 PM

Why do you need to overclock your QX6700, it should already run plenty fast...especially if you actually utilize all 4 cores.
December 25, 2006 5:30:49 AM

Quote:
Its a shame you spent out so much on a CPU and dont have the motherboard to take it to its limit!

Heres another link
http://www.legitreviews.com/article/395/4/

Stable at 3.667ghz with a 1333mhz fsb on 11x on watercooling. Yes please!

Actually I do have the motherboard.
We were talking about someone else’s dilemma.
I was just curious why the P5W DH is not cooperating and was going to see for myself after New Years
!